Generativism and Emergentism: Evidence From Second Language Acquisition Studies of Poverty of the Stimulus Phenomena

Abstract

This work applies evidence from second language acquisition studies to the question of the existence of an inborn, language-specific system that constrains the set of possible human grammars. Three sets of studies are examined and critiqued. The sets were selected so that the studies within each of the sets are comparable on the basis of the target phenomenon investigated and the experimental techniques, but contrast on the basis of their chosen theoretical frameworks. The target phenomena are the Overt Pronoun Constraint, scrambling, and quantifier scope ambiguities. Each set of studies includes at least one study conducted within the generativist approach, advocating for L2 learner access to an innate language-specific endowment (often called Universal Grammar) that is essential to the acquisition of grammar in the presence of impoverished linguistic input. Also, each set of studies contains at least one study conducted within an emergentist approach, advocating that domain-general learning mechanisms applied to rich and complex linguistic input suffice to explain L2 learner attainment. All of the target phenomena investigated in these studies are constructions that have been argued in the generative literature to pose poverty-of-the-stimulus (POS) problems for L2 learners. That is, the linguistic input typically available to learners in the form of second language input, instruction, and through their first language, has been argued to underdetermine the acquisition of the construction. The poverty of the stimulus is considered a cornerstone argument for the existence of UG. Based on my review and critique of the selected studies, I find that their aggregate empirical results support the generativist approach. However, I do not rest this claim on a POS argument; I argue that the premise of insufficiency of input has not been well enough established for the POS argument to be applied. Instead, I argue that the evidence from these studies is more consistent with generative predictions for sudden, step-wise acquisition of grammatical contrasts, possibly on the basis of limited input, rather than with emergentist predictions for gradual, incremental acquisition on a necessarily substantial amount of input.